An invoice arrives at your business email address. Another system sends payment reminders on your behalf. Both use your company’s name, but they need not use the same email provider. Receiving messages and sending messages are different jobs. A DNS report can help explain them, provided we do not mistake one job for the other. 4 6

DNS, the Domain Name System, publishes information associated with names. For email it can publish where incoming messages should go and information that helps recipients check a sender’s use of a domain. Those are not the same record, and neither decides whether a bill should be paid. 2 4 15

First, find the receiving service

In the example address [email protected], the part after @ identifies the recipient’s domain. A sending mail system looks up its mail-delivery instructions. The DNS record used for this is MX, short for “mail exchange.” It names a mail server: a computer system that receives email for the domain. The sender then needs that server’s network address to contact it. 4

Follow a customer’s message step by step. The customer writes an email in a mail program and submits it to their sending service. The service reads the recipient domain, bakery.example, from [email protected]. It asks DNS which mail service receives for that domain. An MX reply might name mail.provider.example. This is a server name, not the bakery’s individual mailbox and not yet a numerical address. The sending service obtains address information for that server name and can then contact the receiving service to offer the message. 4

Only the mail service processes the actual delivery. DNS does not receive the invoice attachment, look inside it or decide which employee opens it. DNS also does not need an MX entry for every staff member: it directs messages for the domain to a receiving service, which handles the recipients. Finding the destination is an earlier step than the receiving service accepting a message or the employee reading it. 4

There can be several MX records. Each has a preference number, and lower numbers are preferred. For example, a destination with preference 10 is normally preferred over one with 20. These are ordering instructions, not percentages of traffic or security scores. Several destinations also need not mean several independent suppliers. 4

This tells us where the bakery’s incoming email is directed. It does not list every service sending mail as the bakery. The accounting system, staff mailboxes and newsletter platform may use different sending arrangements. Nor does an MX record say whether a particular mailbox password is strong. Its job is delivery routing, not a description of all email security. 4 6

One exception is worth understanding before reading a warning about “missing MX.” If the DNS lookup returns no MX records, the mail protocol provides a fallback using the domain’s own address records. This does not promise successful delivery, and it is not a workaround for existing but unusable MX records. A domain deliberately accepting no mail can instead publish an explicit “no mail here” record, called null MX, written as 0 .. A failed lookup, an empty answer and null MX mean different things. 4 5

A TXT record is a place to publish text

Another record type, TXT, stores text. Think of it as a noticeboard: what matters is the notice, not merely the existence of the board. A TXT record might contain information for an email check or something unrelated, such as a service’s verification string. Counting TXT records cannot tell you how well email is protected. 2

A program can read that published text at a specified name and apply rules defined for the kind of notice it recognizes. The text does not execute itself. For example, publishing a sending policy does not make the DNS operator inspect every message; it gives a receiving mail service information for its own checks. Publishing unrelated verification text at the main domain is not the same as publishing a mail policy at the location that a receiver will actually consult. 2 6 15

Now follow a reminder sent by the bakery’s accounting system. The receiving service wants to check its use of the bakery’s domain. Three protections help with different parts of that check.

SPF checks the sending computer. It asks whether that computer is allowed to use a particular domain during mail delivery. The domain being checked comes from the return address for delivery failures or from the sending system’s introduction. It may differ from the address you see beside “From”. So an SPF pass alone does not verify the visible sender address. 6

DKIM checks a message’s signed parts. A message can carry a calculated checking value called a digital signature. DKIM uses that signature for its check. The sending system calculates the signature from those parts using a secret key. A receiving system uses a public key from DNS to check it. A change that alters what was signed prevents that check from succeeding. Think of these as two related pieces of information: one is kept secret for signing, the other is published for checking. Publishing the checking key is intentional, not a password leak. 7

The important difference is between permission about the sending connection and a check attached to a particular message. SPF compares the connecting sender with the published permission. DKIM lets the receiver check the message’s signed parts using the published checking key. The secret signing key is not sent through DNS. If a signature cannot be checked, adding an arbitrary public key somewhere in DNS will not fix it: the key must be the one corresponding to the signer and found where the signature directs the receiver. 6 7

The signature also tells the receiver where to find that public key. It supplies the signing domain and a key label called a selector. A tool that guesses the wrong selector may find nothing even when DKIM is in use. 7

DMARC connects a successful check to the domain shown in “From”. The receiver first needs a successful SPF or DKIM result. It then checks whether the domain used in that successful result matches the visible sender’s domain under DMARC’s rules. This required relationship is called alignment. A pass for an unrelated domain is not enough.

For the reminder in our example, the SPF result used by DMARC concerns the delivery return-address domain. A separate successful SPF check of the sending system’s introduction is not the result DMARC uses here. 15

The domain holder can also publish how it wants receivers to treat messages that fail DMARC. The receiving service makes the final handling decision. These rules help with domain checks; they do not assess the payment instructions inside the message. 15

For example, a reminder could have a valid signature made for an unrelated sending platform’s domain. That result alone does not connect it to the bakery name in “From”. DMARC requires the successful check to have the appropriate relationship to the visible sender’s domain. If the bakery uses an accounting supplier, its administrator and that supplier need to arrange the sending identities and signing appropriately; copying a policy into DNS without arranging the sending side can leave genuine reminders failing checks. 15

An authenticated message is not an approved payment

Suppose an email says that a supplier has changed its bank account. Even a successful DMARC result does not establish that the instruction is honest or commercially authorized. Domain authentication concerns use of a domain, not the truth of every sentence in the message. The technical result cannot replace a payment decision. 15

A practical business rule is to confirm sensitive changes through an already established contact route rather than the new contact details in the request. That advice fits alongside email authentication, not instead of it. One reduces opportunities to misuse your domain; the other helps your staff decide whether an instruction makes sense.

For your own outgoing mail, remember the accounting system as well as staff mailboxes. Before your administrator tightens a policy, explain which services legitimately send messages for the business. An authorized test with an ordinary invented message can then help check each sending path without sharing a real invoice. The useful outcome is working, understandable email—not simply another record added to DNS.

Sources

4 — RFC 5321 — Simple Mail Transfer Protocol

6 — RFC 7208 — Sender Policy Framework

2 — RFC 1035 — Domain names: implementation and specification

15 — RFC 9989 — Domain-Based Message Authentication, Reporting, and Conformance

5 — RFC 7505 — A Null MX Resource Record for Domains That Accept No Mail

7 — RFC 6376 — DomainKeys Identified Mail Signatures